dtody at nrao.edu
Tue Sep 11 14:41:43 PDT 2007
On Tue, 11 Sep 2007, Tony Linde wrote:
> > To expand on this, the VO should only adopt a technology if it brings
> I'd turn it around (more forceful rephrasing of Matthew's) so that the VO
> should always use existing standards unless there is a really good reason to
> invent its own.
Sure, but only if the standard in question first meets *our* requirements
for *our* software, applications, and use-cases. In the case here there
probably is no question that if we wish to experiment with sophisticated
semantic inference applications we should leverage existing technology,
however it may be gross overkill for simple applications, although both
may share the same vocabulary. This matters; a too-complex solution to
a simple problem may not be used (by the broader community or end user)
regardless if it is based on some external standard.
A real-world example of this from outside astronomy is open source
software - why do those guys keep developing new software/standards,
when so much more-or-less relevant stuff already exists? Because it is
not a sufficiently good match for the problem they are trying to solve.
As soon as the product becomes too complex, or clunky, the users won't
use it and they go elsewhere (if they have a choice).
More information about the semantics